KR102671569B1 - Method for providing training content based on ai management provider - Google Patents

Abstract

A method of providing educational content based on an artificial intelligence management provider performed by a main server according to a first embodiment of the present invention, comprising: (a) receiving a question input by a user terminal; (b) the main server classifies the categories of the question and transmits them to the most appropriate learning model among a plurality of learning models to output an answer to the question; and (c) providing the outputted answer to the user terminal, wherein each learning model is trained with learning data consisting of pairs of different question data and answer data according to the category, and Different characters are matched, and when providing the answer, the user terminal also provides an image of the character corresponding to the learning model that output the answer.

Description

Method of providing educational content based on artificial intelligence management provider {METHOD FOR PROVIDING TRAINING CONTENT BASED ON AI MANAGEMENT PROVIDER}

The present invention relates to a method of providing educational content based on an artificial intelligence management provider. More specifically, an artificial intelligence learning model learns pre-entered questions and answers, generates the learned questions and answers into a plurality of answer models, and , relates to a method of providing educational content that classifies categories for questions input by a user terminal and extracts and provides answers from an answer model corresponding to the categories.

As artificial intelligence (AI) technology develops, research is underway to utilize AI learning models in various fields.

In particular, in the field of education that provides educational content to users, research is being conducted most actively due to the advantage of being able to provide educational content to multiple students using an AI learning model trained by a single educator.

In particular, among educational contents using AI learning models, question-and-answer content, in which a user (student) asks a question and an educator (AI) answers the question, is receiving attention as it is judged to be the most suitable.

In this regard, conventional question-and-answer education has been carried out simply by storing expected questions or similar questions in advance on the educator's side, comparing them with questions received from students, and presenting a set answer.

However, this method has the problem that the student's answers are not always consistent, and when written in a different context or grammar, unclear or incorrect answers are presented, reducing the efficiency of question-and-answer learning.

Republic of Korea Patent Publication No. 10-2022-0154935 (Question and answer server control method using artificial intelligence model (publication date: 2022.11.22))

The present invention is intended to solve the problems of the prior art described above, and generates a plurality of answer models from questions and answers previously learned by an artificial intelligence learning model, and extracts and provides answers to questions input by the user terminal from the answer models. The purpose is to

However, the technical challenge that this embodiment aims to achieve is not limited to the technical challenges described above, and other technical challenges may exist.

As a technical means for achieving the above-described technical problem, in the method of providing educational content based on an artificial intelligence management provider performed by the main server according to the first embodiment of the present invention, (a) questions input by the user terminal receiving; (b) the main server classifies the categories of the question and transmits them to the most appropriate learning model among a plurality of learning models to output an answer to the question; and (c) providing the outputted answer to the user terminal, wherein each learning model is trained with learning data consisting of pairs of different question data and answer data according to the category, and Different characters are matched, and when providing the answer, the user terminal also provides an image of the character corresponding to the learning model that output the answer.

In addition, before step (a), collecting learning data for each category and transmitting it to an external server to train the plurality of learning models through the learning data, wherein the external server includes the plurality of learning models. A model is built, and when question data of a specific category is input from the main server, the external server uses the plurality of learning models to output answer data paired with the question data by a learning model corresponding to the specific category. It is a learning server.

In addition, step (b) includes (b-1) the main server classifying categories for the question through a previously learned question classification model; (b-2) transmitting the question and the category value to the external server; (b-3) the external server determines which of the plurality of learning models to input the question into by referring to the category value, inputting the question into the determined model and outputting an answer; and (b-4) transmitting the output answer to the main server.

Additionally, the above categories include how to use educational content, general conversation, English writing, and organizing meeting minutes and class notes.

In addition, the learning model for how to use the educational content receives questions about how to use the educational content through the user terminal and provides answers, and the learning model for the general conversation receives study methods, passing notes, and passing notes from the user terminal. receiving questions about learning, including providing answers corresponding to the questions, and the learning model for English writing receives questions about English writing from the user terminal and provides English writing classes and English writing correction classes. to provide.

In addition, the learning model for English composition provides a ranking of the English composition of the user terminal among all user terminals that performed the English composition with respect to the results of the English composition class and the English composition correction class, and provides a ranking of the English composition of the user terminal. The ranking is determined based on how many words the learning model corrected for the sentence input by the user terminal and how many writing grammar review requests the user terminal input.

In addition, the learning model for English writing is such that when the user terminal repeatedly misspelles a specific word or uses a specific grammar more than a preset number of times during the English writing class and the English writing correction class, the external server sends the information to the user terminal. It is determined that study method guidance is necessary, and information about the current state is transmitted to the learning model for the study method guidance. The learning model for the study method guidance receives information about the current state as an input value, and the specific The answer regarding the spelling of a word or how to study a specific grammar is output as an output value and transmitted to the user terminal.

In addition, the answer UI includes 1) a question window for entering questions, 2) an answer window where answers are provided, and 3) a character display area that displays characters corresponding to learning models that fit the subject of the questions and answers. When entering a question, the answer window and character display area are not provided, but when an answer is provided after entering a question, the answer window and character display area are displayed together.

In addition, a description of the image is received from the user terminal in response to the question, an image generation model corresponding to the learning model generates an image corresponding to the description of the image, and the generated image is sent to the user terminal. If the image is different from the description of the image, the user terminal requests modification of the image.

In addition, different question data and answer data for each specific character are input from the administrator terminal to each learning model to perform learning, and each specific character is taught to provide answers with different vocabulary or tone, and the vocabulary or tone of the answer is learned. If (i.e., style) is different from the target, feedback on the answer is input to the learning model and re-training is performed.

In addition, the learning model for the general conversation analyzes the questions of the user terminal to collect words included in the user terminal's questions, distinguishes frequently used words to determine the field of interest of the user terminal, and determines the user terminal's field of interest. Answers to questions on the terminal are provided according to the above-mentioned areas of interest.

In addition, after step (c), if the user terminal repeatedly inputs the question more than a preset number of times, a notification is sent to the administrator terminal to check whether the answer to the question of the user terminal is incorrect. .

In addition, when questions are entered into the learning model, Creativity, Collaboration, Critical Thinking, Compassion, Sincerity, Passion, Global Citizenship Train a learning model that outputs a result as one of eight tendencies including citizenship and friendship, and after converting the question into a vector, the vocabulary of words or concluding sentences converted into vectors is trained. A weight for each of the eight tendencies is set, and when a specific question is input, learning is performed on the weight based on a supervised learning model or reinforcement learning model so that any one of the eight tendencies is output. As the output value, Provides one of eight tendencies and the level value for that tendency as a number, calculates the average level value for the tendency of the user's questions during a specific period, and calculates the average level for the tendency among the user's questions during a specific period. By excluding and analyzing questions that differ from the level value, the user's tendency analysis is completed.

In addition, in response to the question input by the user terminal, the main server provides an answer output from the external server and a positive or negative button for each answer, and when positive feedback corresponding to the positive button is received. , the learning model is retrained so that an answer corresponding to the positive feedback is provided when the same question is inputted later, and when negative feedback corresponding to the negative button is received, the learning model provides a different answer and is re-trained. Perform learning, set the score and category for each answer according to the user's feedback, store the history of the user's feedback, and later, when a question is input from the user terminal, the score and category for each answer Based on the category and the history, the learning model provides answers tailored to the user's preferences.

An apparatus for providing educational content based on an artificial intelligence management provider performed by a main server, comprising: a memory storing a program for a method of providing educational content based on an artificial intelligence management provider; and a processor for executing the program, the method comprising: (a) receiving a question input by a user terminal; (b) the main server classifies the categories of the question and transmits them to the most appropriate learning model among a plurality of learning models to output an answer to the question; and (c) providing the outputted answer to the user terminal, wherein each learning model is trained with learning data consisting of pairs of different question data and answer data according to the category, and Different characters are matched, and when providing the answer, the user terminal also provides an image of the character corresponding to the learning model that output the answer.

The present invention can generate questions and answers previously learned by an artificial intelligence learning model into a plurality of answer models, and extract and provide answers to questions input by the user terminal from the answer models.

1 is a structural diagram of a method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.
Figure 2 is a structural diagram of the main server of the method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.
Figure 3 is an example diagram showing the structure of an external server, a learning model, and characters in a method of providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.
4A to 4B are exemplary diagrams showing the answer UI of the method of providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.
Figure 5 is a flowchart showing the process of a method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.
Figure 6 is an exemplary diagram showing an image creation model of an artificial intelligence management provider-based method of providing educational content according to an embodiment of the present invention.
Figure 7 is an example diagram showing a study sharing and question and answer area provided during learning of an artificial intelligence management provider-based educational content provision method according to an embodiment of the present invention.

Below, with reference to the attached drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily implement the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. In order to clearly explain the present invention in the drawings, parts unrelated to the description are omitted, and similar parts are given similar reference numerals throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected," but also the case where it is "electrically connected" with another element in between. . Additionally, when a part "includes" a certain component, this means that it may further include other components rather than excluding other components, unless specifically stated to the contrary.

In this specification, 'part' includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Additionally, one unit may be realized using two or more pieces of hardware, and two or more units may be realized using one piece of hardware. Meanwhile, '~ part' is not limited to software or hardware, and '~ part' may be configured to reside in an addressable storage medium or may be configured to reproduce one or more processors. Therefore, as an example, '~ part' refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and 'parts' may be combined into a smaller number of components and 'parts' or may be further separated into additional components and 'parts'. Additionally, components and 'parts' may be implemented to regenerate one or more CPUs within a device or a secure multimedia card.

The “user terminal” mentioned below may be implemented as a computer or portable terminal that can connect to a server or other terminal through a network. Here, the computer is, for example, a laptop equipped with a web browser, a desktop, a laptop, a VR HMD (e.g., HTC VIVE, Oculus Rift, GearVR, DayDream, PSVR, etc.), etc. may include. Here, VR HMD is for PC (e.g. HTC VIVE, Oculus Rift, FOVE, Deepon, etc.), mobile (e.g. GearVR, DayDream, Storm Magic, Google Cardboard, etc.), and console (PSVR). Includes independently implemented Stand Alone models (e.g. Deepon, PICO, etc.). Portable terminals are, for example, wireless communication devices that ensure portability and mobility, including smart phones, tablet PCs, and wearable devices, as well as Bluetooth (BLE, Bluetooth Low Energy), NFC, RFID, and ultrasonic devices. , may include various devices equipped with communication modules such as infrared, WiFi, and LiFi. In addition, “network” refers to a connection structure that allows information exchange between nodes such as terminals and servers, including a local area network (LAN), a wide area network (WAN), and the Internet. (WWW: World Wide Web), wired and wireless data communication network, telephone network, wired and wireless television communication network, etc. Examples of wireless data communication networks include 3G, 4G, 5G, 3GPP (3rd Generation Partnership Project), LTE (Long Term Evolution), WIMAX (World Interoperability for Microwave Access), Wi-Fi, Bluetooth communication, infrared communication, and ultrasound. This includes, but is not limited to, communication, Visible Light Communication (VLC), LiFi, etc.

First, with reference to FIG. 1, the structure of a method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention will be described.

According to Figure 1, the method of providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention includes a main server 100, an external server 200, a user terminal 300, and an administrator terminal 400. . However, the structure of the present invention is not limited to this.

The main server 100 may receive data on questions and answers from the administrator terminal 400, perform learning, and receive questions in the form of text or images from the user terminal 300. Hereinafter, the structure of the main server 100 will be described with reference to FIG. 2.

Referring to Figure 2, the main server 100 of the artificial intelligence management provider-based educational content provision method according to an embodiment of the present invention may include a memory 110, a processor 120, and a communication module 130. there is. However, the structure of the main server 100 of the present invention is not limited to this.

The memory 110 stores at least one of information and data input to the communication module 130, information and data required for functions performed by the processor 120, and data generated according to execution of the processor 120. You can. Memory 110 should be interpreted as a general term for non-volatile storage devices that continue to retain stored information even when power is not supplied and volatile storage devices that require power to maintain stored information. Additionally, the memory 110 may perform a function of temporarily or permanently storing data processed by the processor 120. The memory 110 may include magnetic storage media or flash storage media in addition to volatile storage devices that require power to maintain stored information, but the scope of the present invention is not limited thereto. no.

The memory 110 stores the educational content providing program, and allows the user terminal 300 to receive information input within the educational content through the processor 120 and link it to the program. The name of the program is set for convenience of explanation, and the name itself does not limit the function of the program.

The processor 120 is configured to execute an educational content providing program stored in the memory 110. The processor 120 may include various types of devices that control and process data. The processor 120 may refer to a data processing device built into hardware that has a physically structured circuit to perform functions expressed by codes or instructions included in a program. In one example, the processor 120 may include a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated circuit (ASIC), or an FPGA ( It may be implemented in the form of a field programmable gate array, etc., but the scope of the present invention is not limited thereto.

The processor 120 is configured to perform the detailed processes of the artificial intelligence management provider-based educational content provision method described below.

The communication module 130 can transmit and receive information with an external database or external device. Here, the external device may be a user terminal 300 and an administrator terminal 400. The communication module 130 may include a device including hardware and software necessary to transmit and receive signals such as control signals or data signals through wired or wireless connections with other network devices.

The communication module 130 transmits the educational content provision program stored in the memory 110 provided by the processor 120 of the server 100 to an external database or external device so that it can be linked.

The external server 200 will be described below with reference to FIG. 3.

Figure 3 is an example diagram showing the structure, learning model, and characters of the external server 200 of the method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.

According to Figure 3, the external server 200 of the present invention can be constructed with a plurality of learning models including a first learning model, a second learning model, a third learning model, and a fourth learning model. However, the structure of the external server 200 is not limited to this. And, when question data of a specific category is input from the main server 100, the external server 200 can learn a plurality of learning models to output answer data paired with the question data by a learning model corresponding to the specific category. . Additionally, different characters 201 may be provided by matching each learning model. When an answer to a question of the user terminal 300 is provided, these characters may also provide an image of the character 201 in response to the learning model that outputs the answer. Each learning model can be one of supervised learning, unsupervised learning, and reinforcement learning models. If it is a supervised learning model, the learning model is created by assigning labeling values to pairs of question data and answer data. You can output the desired output value.

The user terminal 300 may include a smartphone, tablet PC, notebook (laptop), and desktop that are configured with a screen and can perform input. However, the user terminal 300 is not limited to this. The user terminal 300 can transmit a question in the form of text or an image to the server 100 and check the answer provided by the server 100.

The administrator terminal 400 may include a smartphone, tablet PC, notebook (laptop), and desktop that are configured with a screen and can perform input. However, the manager terminal 400 is not limited to this. The administrator terminal 400 inputs a plurality of questions and answers into the server 100.

4A to 4B are exemplary diagrams showing the answer UI of the method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention.

According to Figure 4a, a question window 500 for entering a question may be displayed in one area of the answer UI of the present invention. However, the position of the question window 500 is not limited as shown in FIG. 4A. The user terminal 300 may input “Tell me whether the sentence “PLEASE GIVES MI BYCICLE” is grammatically correct” as shown in FIG. 4A in the provided question window 500. Then, the question can be transmitted by entering the transmit button placed in the first area of the question window 500.

According to Figure 4b, in one area of the answer UI of the present invention, there is a question window 500 for entering a question, an answer window 600 where an answer is provided, and a character 201 corresponding to a learning model suitable for the topic of the question and answer. ) can be displayed in the character display area. However, the positions of the question window 500, the answer window 600, and the character display area are not limited to this. In Figure 4a, when the user terminal 300 inputs a question and presses the send button, as shown in Figure 4b, the main server 100 sends "PLEASE GIVE ME A BICYCLE." output from the external server 200 to the user terminal 300. It is sent to the reply window (600). At this time, a character 201 is displayed adjacent to the answer window 600.

At this time, when entering a question in the question window 500 as shown in Figure 4a, the answer window 600 and the character display area are not provided, but when an answer is provided after entering the question, the answer window 600 and the character The display area may be displayed together.

Therefore, in the answer window 600, the question entered in the question window 500 by the user terminal 300, the answer extracted from the external server 200 through the main server 100, and a character matching the learning model containing the answer ( 201) may be displayed together.

Hereinafter, with reference to FIG. 5, a sequence showing the process of the method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention will be described.

First, the manager terminal 400 inputs question data and answer data for each category into the main server 100 (S510). Next, the question data and answer data for each category collected by the main server 100 are input into the AI engine of the external server 200 to perform learning (S520). Next, the external server 200 builds the first learning model, the second learning model, the third learning model, and the fourth learning model (S530).

The main server 100 can collect question data and answer data by category from the administrator terminal 400. Here, question data and answer data consisting of pairs can be referred to as learning data. By transmitting the collected training data for each category to the external server 200, learning can be performed by matching the training data to a learning model corresponding to the category among a plurality of learning models. At this time, the category may include how to use educational content, general conversation, English writing, organizing meeting minutes and class notes, but the category is not limited to these.

If explained by category, the learning model for how to use educational content receives questions about how to use educational content through the user terminal 300 and provides answers, and the learning model for general conversation receives study methods and passing notes from the user terminal 300. and answers are provided by receiving questions about learning, including passing notes, and the learning model for English writing receives questions about English writing from the user terminal 300 and provides English writing classes and English writing correction classes. can do.

Next, the main server 100 receives the question input by the user terminal 300 (S540). As described in FIG. 4 above, the user terminal 300 can input a question into the question window 500 of the answer UI provided to the user terminal 300 and transmit it to the main server 100.

Next, the main server 100 classifies the questions into categories and transmits the questions and categories to the external server 200 (S550). The main server 100 may recognize words included in the question received from the user terminal 300, classify them into categories, and transmit the question and category to the external server 200.

Next, the external server 200 determines which of the plurality of answer models to input the question into by referring to the category, inputs the question into the determined model, and outputs the answer (S560). The external server 200 may determine which of the plurality of learning models to input the question into by referring to the category value, input the question into the determined model, and output an answer. The printed answer can be transmitted to the main server (100).

Next, the answer output by the external server 200 is provided to the user terminal 300 (S570). The main server 100 may receive the answer output from the external server 200 and provide it to the user terminal 300.

Hereinafter, with reference to FIG. 6, the image generation model of the method for providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention will be described.

According to FIG. 6, in steps S510 to S570 of FIG. 5 of the present invention, ① a description of the imagined image is received from the user terminal 300 in response to a question, and ② an image generation model corresponding to the learning model provides a description of the image. An image corresponding to can be generated and the generated image can be provided to the user terminal 300. Although not shown in FIG. 6, if the image generated by the image generation model is different from the description of the image, the user terminal 300 may request modification of the image. And, if an image identical to the image described by the user terminal 300 is generated, the image may be provided to another user terminal 300.

Hereinafter, with reference to FIG. 7, the study sharing and question-and-answer areas provided during learning of the method of providing educational content based on an artificial intelligence management provider according to an embodiment of the present invention will be described.

The answer UI of the present invention described above with reference to FIGS. 4A and 4B is a screen provided for questions and answers of the user terminal 300, and FIG. 7 shows a screen provided for questions and answers while the user terminal 300 is learning. It's a screen. According to FIG. 7 , the screen provided while the user terminal 300 is learning may include a study sharing area 301 and a question and answer area 302. In the study sharing area 301, study screens of other user terminals including CAM1, CAM2, ??, CAM9, ??, CAMn are shared, and a question window 500 is provided in the question and answer area 302. If a question arises while performing learning, the user terminal 300 inputs a question into the question window 500 of the question and answer area 302 and transmits it, and the main server 100 sends the answer output from the external server 200 to the user terminal. It can be provided as (300).

As a first additional embodiment of the present invention, the learning model for English composition is the ranking of the English composition of the user terminal 300 among all user terminals 300 that performed English composition as a result of the English composition class and the English writing correction class. can be provided together. The writing ranking will be determined based on how many words the learning model corrected for the sentence entered by the user terminal 300 and how many writing grammar review requests the user terminal 300 entered. You can.

As a second additional embodiment of the present invention, a learning model for English writing is provided when the user terminal 300 repeatedly makes a mistake in spelling a specific word or specific grammar more than a preset number of times during English writing class and English writing correction class. The server 200 determines that the user terminal 300 needs study method guidance, and provides information about the current status as a learning model for study method guidance (e.g., OO word spelled incorrectly more than 5 times, or TO infinitive) You can convey grammar mistakes more than 7 times. The learning model for teaching study methods receives information about the current state as an input value, outputs an answer regarding a study method that can solve the problem as an output value, and transmits it to the user terminal 300. In the user terminal 300, a character related to grammar correction appears and an answer is provided. If the same grammar or word spelling is detected as incorrect, another character appears and an answer providing advice on study methods may be provided. .

As a third additional embodiment of the present invention, a character to be used by the user terminal 300 from among a plurality of characters provided can be selected or customized to be created. For example, in order for the user terminal 300 to select one of predetermined characters consisting of a man, a woman, an animal, etc. or to create its own character, the shape of the face, shape and color of the head, shape and color of the eyes, nose, etc. You can customize it by determining the shape of the mouth, shape of the mouth, accessories, etc.

As a fourth additional embodiment of the present invention, the administrator terminal 400 can be taught to provide answers with different vocabulary or tone for each specific character. To this end, learning can be performed by inputting different question data and answer data into each learning model. If the vocabulary or tone (i.e. style) of the answer is different from the target, feedback on the answer can be input into the learning model and re-training can be performed. According to this iterative learning, the user terminal 300 can receive answers according to the character's answer style from a character whose personality matches the category of the question.

For example, when the user terminal 300 asks a daily question or a learning method, an answer can be provided in a friendly tone, as if a friend or senior around you explains it, or when the user terminal 300 asks a question about English writing or English writing correction. , answers can be provided in the language of a teacher who teaches learning, such as suggesting meticulous and accurate writing and correction, and informing students of expressions that are prone to mistakes.

As a fifth additional embodiment of the present invention, the learning model for general conversation analyzes the questions of the user terminal 300, collects words included in the questions of the user terminal 300, and distinguishes frequently used words from the user terminal 300. It is possible to determine the field of interest of the user terminal 300 and provide answers to questions of the user terminal 300 according to the field of interest.

As a sixth additional embodiment of the present invention, when the user terminal 300 repeatedly inputs a question more than a preset number of times, the administrator terminal 400 is notified to check whether the answer to the question of the user terminal 300 is incorrect. can be transmitted. Because the answer to the question entered by the user terminal 300 may not appear correctly, the administrator terminal 400 can check and modify the answer.

As an eighth additional embodiment of the present invention, the learning model for English writing can search for and provide sentences similar to the question on the Internet when the answer corresponding to the question of the user terminal 300 does not exist in the learning data. At this time, a learning model for English writing determines whether the sentence searched on the Internet contains grammatical errors and provides it.

As a seventh additional embodiment of the present invention, when the user terminal 300 inputs a sentence that is not a question, the main server 100 provides another answer corresponding to the sentence or asks the user terminal 300 a desired question from the sentence. It is judged and provided in reverse.

For example, if the user terminal 300 inputs “PLEASE GIVES MI BICICLE,” it may provide a different answer, such as “That is not the correct sentence,” rather than correcting the sentence. Accordingly, if the user terminal 300 inputs “Tell me the correct grammar,” it can provide the answer “PLEASE GIVE ME A BICYCLE.”

Or, when you enter "PLEASE GIVES MI BICICLE", you can ask the reverse question, "Can you tell me the correct grammar for "PLEASE GIVES MI BICICLE"?" At this time, if the user terminal 300 inputs “Yes” or “Yes,” it may provide the answer “PLEASE GIVE ME A BICYCLE.”

As an eighth additional embodiment of the present invention, the user's tendency to input questions through the user terminal 300 can be analyzed. For example, when questions are entered into the preset learning model, Creativity, Collaboration, Critical Thinking, Compassion, Sincerity, Passion, Global You can train a learning model that outputs results in one of eight tendencies: Citizenship (global citizenship) and Friendship (relationship). For example, after converting the user's question into a vector, the words or concluding words converted into vectors are classified into Creativity, Collaboration, Critical Thinking, Compassion, and Sincerity. ), Passion (passion), Global Citizenship (global citizenship), and Friendship (relationship) are set as weights, and these weights can be changed by the learning model itself to perform learning. Supervised learning models and reinforcement learning models can be used here. Additionally, the level value for each user's tendency can be expressed numerically. The external server's learning model provides the printed user's tendency and level value to the main server, and the main server can provide this to the user terminal. Users will be able to objectively understand their tendencies by checking these indicators. The main server can calculate the average level of questions generated by users over a specific period of time. The main server extracts questions with a level that is different from the average level of user questions created during a specific period by more than a preset value, and excludes the questions from the data for user tendency analysis. Accordingly, the main server can increase the accuracy of user preference analysis. Afterwards, the main server can complete the analysis of the user's tendencies for a specific period of time, search the database for mentors (experts) that match the user's tendencies, and provide them to the user terminal. For example, if the user's tendency is determined to have a value of level 4 for Passion, academy instructors who have received passionate teaching evaluations from the list of academy instructors can be extracted as a mentor list and provided to the user terminal.

As a ninth additional embodiment of the present invention, the user terminal can perform a general conversation, and at this time, the artificial intelligence learning model (learning model for general conversation) can be trained to have a conversation in a way that matches the user's inclination. For example, for the input sentence “I’m hungry,” the learning model provides 1) Problem-solving response: “Then go to a restaurant and eat.” 2) Experience-sharing response: “I’m hungry, too.” 3) Emotional response: “I’m very hungry.” You can answer one of the following: “I’m so hungry that I want to quit studying right away.” However, since each user will have different preferences and friends will also be different, the preferences of the answers each user wants to receive will also be different. Users can train the learning model to provide answers that suit their preferences. For example, after entering a specific input sentence, if you like the output sentence output from an external server and delivered through the main server, you can provide positive feedback to the learning model by entering the positive button next to the dialog window. there is. However, if the user does not like the output sentence, he or she can provide negative feedback by entering the negative button, and the learning model can provide a different answer again. At this time, the external server can set identification information for each user terminal and store in advance which answer is preferred for each identification information based on input to the positive and negative buttons. Afterwards, when the user asks a specific question, the learning model outputs multiple answers, sets the score and category (emotional type, problem-solving type, experience sharing type, etc.) for each answer, and allows the user to previously input By referring to the history of the positive and negative buttons, you can select the answer that the user is presumed to prefer the most. During this process, if the user inputs "In that case, you would just answer that you were very upset," the learning model recognizes that the phrase "You would answer ~" in the sentence as indicating a specific answer, and the user You can remember the answer you want.

One embodiment of the present invention may also be implemented in the form of a recording medium containing instructions executable by a computer, such as program modules executed by a computer. Computer-readable media can be any available media that can be accessed by a computer and includes both volatile and non-volatile media, removable and non-removable media. Additionally, computer-readable media may include all computer storage media. Computer storage media includes both volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data.

Although the methods and systems of the present invention have been described with respect to specific embodiments, some or all of their components or operations may be implemented using a computer system having a general-purpose hardware architecture.

The description of the present invention described above is for illustrative purposes, and those skilled in the art will understand that the present invention can be easily modified into other specific forms without changing the technical idea or essential features of the present invention. will be. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. For example, each component described as single may be implemented in a distributed manner, and similarly, components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. do.

100: main server 110: memory
120: processor 130: communication module
200: External server 201: Character
300: User terminal 301: Study sharing area
302: Question and answer area 400: Administrator terminal
500: Question window 600: Answer window

Claims (15)

In the method of providing educational content based on an artificial intelligence management provider performed by the main server,
(a) receiving a question input by a user terminal;
(b) the main server classifies the categories of the question and transmits them to the most appropriate learning model among a plurality of learning models to output an answer to the question; and
(c) providing the printed answer to the user terminal;
Includes,
Each learning model is trained with learning data consisting of pairs of different question data and answer data according to the above categories, and different characters are matched,
When providing the answer, the user terminal also provides an image of a character corresponding to the learning model that output the answer,
The above categories include how to use educational content, general conversation, English writing, organizing meeting minutes and class notes,
The learning model for English writing above is
During the English writing class and English writing correction class, if the user terminal repeatedly misspelled a specific word or specific grammar more than a preset number of times,
The external server determines that the user terminal needs study method guidance,
It delivers information about the current status as a learning model for guiding the study method,
The learning model for teaching the study method receives information about the current state as input,
An answer regarding the spelling of the specific word or a study method of a specific grammar is output as an output value and transmitted to the user terminal,
Receiving a description of the image from the user terminal in response to the question,
An image generation model corresponding to the learning model generates an image corresponding to a description of the image,
Providing the generated image to the user terminal,
If the image is different from the description of the image, the user terminal requests modification of the image,
Different question data and answer data for each specific character from the administrator terminal are input into each learning model to perform learning,
Learn to provide answers using different vocabulary or tone for each specific character,
If the vocabulary or tone (i.e. style) of the answer is different from the target, feedback on the answer is input into the learning model and re-training is performed.
The learning model for the general conversation analyzes the questions of the user terminal, collects words included in the questions of the user terminal, and determines the field of interest of the user terminal by distinguishing frequently used words,
Provide answers to questions from the user terminal according to the area of interest,
When the user terminal repeatedly inputs the question more than a preset number of times,
Sending a notification to the administrator terminal to check whether the answer to the question of the user terminal is incorrect,
When questions are entered into the learning model above, Creativity, Collaboration, Critical Thinking, Compassion, Sincerity, Passion, Global Citizenship ), train a learning model that outputs results as one of eight tendencies including Friendship (relationship),
After converting the question into a vector, a weight for each of the eight tendencies is set for the vocabulary of words or endings converted into vectors, and when a specific question is input, the weight is set so that one of the eight tendencies is output. Performs learning based on a supervised learning model or reinforcement learning model,
As the output value, one of the eight tendencies and the level value for that tendency are provided as numbers,
Completing the user's tendency analysis by calculating the average level value of the user's question tendency during a specific period and excluding and analyzing questions that differ from the average level value of the user's question during a certain period of time. do,
In response to the question input by the user terminal, the main server provides an answer output from the external server and a positive or negative button for each answer,
When positive feedback corresponding to the positive button is received, the learning model is retrained so that an answer corresponding to the positive feedback is provided when the same question is input later,
When negative feedback corresponding to the negative button is received, the learning model provides a different answer and performs re-learning,
According to the user's feedback, the score and category for each answer are set, and the history of the user's feedback is stored,
Afterwards, when a question is input from the user terminal, the learning model provides an answer tailored to the user's preference based on the score and category for each answer and the history,
The learning model for English writing is
If an answer corresponding to the question of the user terminal does not exist in the learning data, a sentence similar to the question is searched on the Internet and provided,
If the grammar of the sentence searched on the Internet is incorrect, the learning model for English writing determines and provides it,
If the user terminal enters a sentence that is not a question,
A method of providing educational content based on an artificial intelligence management provider, which provides another answer corresponding to the sentence or determines the question desired by the user terminal from the sentence and provides the reverse answer. According to claim 1,
Before step (a) above
Collecting learning data for each category, transmitting it to an external server, and training the plurality of learning models through the learning data;
Including,
The plurality of learning models are built on the external server, and when question data of a specific category is input from the main server, the external server generates answer data paired with the question data by a learning model corresponding to the specific category. A method of providing educational content based on an artificial intelligence management provider, which is a server that learns the plurality of learning models to output. According to claim 1,
In step (b),
(b-1) the main server classifying categories for the question through a previously learned question classification model;
(b-2) transmitting the question and the category value to the external server;
(b-3) the external server determines which of the plurality of learning models to input the question into by referring to the category value, inputting the question into the determined model and outputting an answer; and
(b-4) transmitting the output answer to the main server;
A method of providing educational content based on an artificial intelligence management provider, further comprising: delete According to claim 1,
The learning model for how to use the educational content receives questions about how to use the educational content through the user terminal and provides answers,
The learning model for the general conversation receives questions about learning including study methods, passing notes, and passing notes from the user terminal, and provides answers corresponding to the questions,
The learning model for English writing receives questions about English writing from the user terminal and provides English writing classes and English writing correction classes. A method of providing educational content based on an artificial intelligence management provider. According to claim 5,
The learning model for English writing above is
About the results of the above English writing class and English writing correction class
Provides a ranking of the English composition of the user terminal among all user terminals that performed the English composition,
The ranking of the composition is determined based on how many words the learning model corrected for the sentence entered by the user terminal and how many composition grammar review request inputs the user terminal entered. , Method of providing educational content based on artificial intelligence management provider. delete According to claim 1,
The answer UI includes 1) a question window for entering questions, 2) an answer window where answers are provided, and 3) a character display area that displays characters corresponding to a learning model that fits the topic of the questions and answers.
When entering a question in the question window, the answer window and character display area are not provided, but when an answer is provided after entering a question, the answer window and character display area are displayed together. Education based on artificial intelligence management provider. How to provide content. delete delete delete delete delete delete In an educational content provision device based on an artificial intelligence management provider performed by the main server,
Memory for storing a program related to an artificial intelligence management provider-based method of providing educational content; and
Includes a processor for executing the program,
The above method is
(a) receiving a question input by a user terminal;
(b) the main server classifies the categories of the question and transmits them to the most appropriate learning model among a plurality of learning models to output an answer to the question; and
(c) providing the printed answer to the user terminal;
Includes,
Each learning model is trained with learning data consisting of pairs of different question data and answer data according to the above categories, and different characters are matched,
When providing the answer, the user terminal also provides an image of a character corresponding to the learning model that output the answer,
The above categories include how to use educational content, general conversation, English writing, organizing meeting minutes and class notes,
The learning model for English writing is
During the English writing class and English writing correction class, if the user terminal repeatedly misspelled a specific word or specific grammar more than a preset number of times,
The external server determines that the user terminal needs study method guidance,
It delivers information about the current status as a learning model for guiding the study method,
The learning model for teaching the study method receives information about the current state as input,
An answer regarding the spelling of the specific word or a study method of a specific grammar is output as an output value and transmitted to the user terminal,
Receiving a description of the image from the user terminal in response to the question,
An image generation model corresponding to the learning model generates an image corresponding to a description of the image,
Providing the generated image to the user terminal,
If the image is different from the description of the image, the user terminal requests modification of the image,
Different question data and answer data for each specific character from the administrator terminal are input into each learning model to perform learning,
Learn to provide answers using different vocabulary or tone for each specific character,
If the vocabulary or tone (i.e. style) of the answer is different from the target, feedback on the answer is input into the learning model and re-training is performed.
The learning model for the general conversation analyzes the questions of the user terminal, collects words included in the questions of the user terminal, and determines the field of interest of the user terminal by distinguishing frequently used words,
Provide answers to questions from the user terminal according to the area of interest,
When the user terminal repeatedly inputs the question more than a preset number of times,
Sending a notification to the administrator terminal to check whether the answer to the question of the user terminal is incorrect,
When questions are entered into the learning model above, Creativity, Collaboration, Critical Thinking, Compassion, Sincerity, Passion, Global Citizenship ), train a learning model that outputs results as one of eight tendencies including Friendship (relationship),
After converting the question into a vector, a weight for each of the eight tendencies is set for the vocabulary of words or endings converted into vectors, and when a specific question is input, the weight is set so that one of the eight tendencies is output. Performs learning based on a supervised learning model or reinforcement learning model,
As the output value, one of the eight tendencies and the level value for that tendency are provided as numbers,
Completing the user's tendency analysis by calculating the average level of the user's question tendency during a specific period and excluding and analyzing questions that differ from the average level value of the user's question during the specific period. do,
In response to the question input by the user terminal, the main server provides an answer output from the external server and a positive or negative button for each answer,
When positive feedback corresponding to the positive button is received, the learning model is retrained so that an answer corresponding to the positive feedback is provided when the same question is input in the future,
When negative feedback corresponding to the negative button is received, the learning model provides a different answer and performs re-learning,
According to the user's feedback, the score and category for each answer are set, and the history of the user's feedback is stored,
Afterwards, when a question is input from the user terminal, the learning model provides an answer tailored to the user's preference based on the score and category for each answer and the history,
The learning model for English writing is
If an answer corresponding to the question of the user terminal does not exist in the learning data, a sentence similar to the question is searched on the Internet and provided,
If the grammar of the sentence searched on the Internet is incorrect, the learning model for English writing determines and provides it,
If the user terminal enters a sentence that is not a question,
An educational content providing device based on an artificial intelligence management provider, which provides another answer corresponding to the sentence or determines a question desired by the user terminal from the sentence and provides the reverse answer.

Images